Decoherence of a Coherent Beam If retarding films (i.e. wave plates) can retard the electric components of light (based on the films thickness) without affecting it's direction, couldn't I use a retarding film of randomly varying thickness to convert a Coherent laser beam into an incoherent laser beam to improve eye safety? 
For example, say I have a 632nm laser source. I make a retarding film that is optically flat on one side, but the other side contains nanoscale smooth bumps between 0 and 635nm tall (above the surface) and I pass the laser through it (perhaps a series of these films), wouldn't an incoherent collimated source be generated. Alternatively, in lieu of bumps, I think you could also use a retarder comprised of randomly oriented liquid crystal molecules in a polymer matrix. 
Thanks.
 A: 
Couldn't I use a retarding film of randomly varying thickness to convert a Coherent laser beam into an incoherent laser beam...?

If the film is not changing (in time) you are not changing the coherence properties of beam at all.  You can think of putting a slab of something in the way as putting a really bad lens (possibly with no optical power) in the path of the beam. If you pick any two points (either at different points in space if you mean spatial coherence or two points in time if you mean temporal coherence) and you take the light from these two points and combine them, you will still see interference. The only thing you have changed is the phase of the fringes has changed (which means the fringes might shift, but you haven't changed the visibility and therefore the coherence of the fringes).
Now if you could produce random phases that were constantly changing faster than your "detector" is "measuring" the light (e.g. if you are thinking of eye safety then this might mean change fast enough that a single coherent "instant" is too fast for your eye to respond and/or become damaged), then your detector will average over the changes and you effectively have partially coherent or incoherent light.
This is actually done in experiments. For example there are experiments that will shine a laser through a ground glass plate that is rotating [Ref 1], or you can quickly (and precisely) change the phases across the beam on a spatial light modulator [Ref 2] (note they use a micro-mirrored array or DMD like those found in projectors, as normal LCD displays are usually not fast enough to do this).


*

*A. Gatti et al. "Coherent imaging with pseudo-thermal incoherent light." Journal of Modern Optics, 53 (2006). doi:10.1080/09500340500147240 arXiv:quant-ph/0504082

*B. Rodenburg et al. "Experimental generation of an optical field with arbitrary spatial coherence properties." JOSA B 31 (2014). doi:10.1364/JOSAB.31.000A51 arXiv:1312.6878
A: 
...couldn't I use a retarding film of randomly varying thickness to convert a Coherent laser beam into an incoherent laser beam to improve eye safety?

Absolutely not. A beam's destructiveness to the eye depends on three things:


*

*Energy delivered to retina and the time periods it is delivered over, quantified by the ISO60825 concept of Maximum Permissable Exposure;

*Spectral content: some wavelengths are short enough to cause photochemical damage: Maximum Permissable Exposure limits in ISO60825 are wavelength dependent;

*Indirectly, in some cases, whether the beam can be seen and thus whether the viewer can thus shield their eyes through the blink reflex. This is the grounds of the laser classes 2 and 3R: a fleeting viewing of the former cut short by the normal blink reflex is deemed safe by ISO60825 and the same glimpse of a 3R beam is of greatly mitigated risk. 


Coherence has nothing to do directly with a light beam's hazard to the eyes. Indeed the laser safety standard ISO60825-2014 makes no distinction between coherent and incoherent sources. You assess the safety of an LED in exactly the same way as you do a laser. A proviso to this statement: if you add aberration or incoherence to a beam in the way you propose, you will change its divergence / convergence behaviour and thus you will change the power density reaching the retina in certain situations. Some coherence properties thus influence, indirectly, the calculation of the Nominal Occular Hazard Zone and Nominal Optical Hazard Distance from sources.
Moreover, to simulate incoherent light as in Punk Physicist's answer, you need to answer exactly what you are simulating incoherence for. To be pedantic, a truly partially coherent beam is a classical mixture of pure photon states, as I describe in my answer here. So, the only way to produce truly incoherent light from coherent light is to randomize each photon's phase / polarization by making it interact somehow with a truly random process: a thermalized system or radioactive source, for example. 
The methods for decohering light in Punk Physicist's answer only produce pseudo-decohered light, which has the same relationship with a truly incoherent / partially coherent source as a pseudo-random number sequence produced by a deterministic numerical algorithm has to a a truly random sequence produced by e.g. a quantum system. This is nonetheless useful for many applications, in the same way that pseudo-random sequences are highly useful.
